Canada

The representation degeneration problem is a phenomenon that is widely observed among self-supervised learning methods based on Transformers. In NLP, it takes the form of anisotropy, a singular property of hidden representations which makes them unexpectedly close to each other in terms of angular distance (cosine-similarity). Some recent works tend to show that anisotropy is a consequence of optimizing the cross-entropy loss on long-tailed distributions of tokens. We show in this paper that anisotropy can also be observed empirically in language models with specific objectives that should not suffer directly from the same consequences. We also show that the anisotropy problem extends to Transformers trained on other modalities. Our observations tend to demonstrate that anisotropy might actually be inherent to Transformers-based models.

ACL 2023

Is Anisotropy Inherent to Transformers?

representation degeneration

anisotropy

transformers

poster

### Welcome to ACL 2023, the 61st Annual Meeting of the Association for Computational Linguistics! 
<br>The conference will be held in Toronto, Canada, July 9-14, 2023.<br><br>
Following the succession of the recent conferences in our field, ACL 2023 will adopt a hybrid format.
While the impact of Covid has considerably diminished in terms of traveling, obtaining visas to Canada
entails a very long process. Moreover, the global economic conditions pose challenges for many individuals to travel to conferences. Recognizing these circumstances, we know many participants may not be
able to attend the conference in person. Therefore, we are committed to providing a great virtual platform
so everyone has the opportunity to interact with other participants and enjoy the conference. Based on the
current registered participants, approxiately 30% have chosen to attend the conference virtually. Whether
you join us in person or virtually, we sincerely hope everyone has a remarkable conference experience.<br>
This General Chair’s message is where I express my gratitude to the many individuals who have made
enormous contributions to the conference over the past year.

Read [**ACL 2023 General Chair's message**](https://docs.google.com/document/d/1WobYM7norbG4dI48s75HfJoD89qgX5a_F-6U8AteLSA/edit?usp=sharing/) in full.

##### **[Conference Handbook](https://2023.aclweb.org/downloads/acl2023-handbook.pdf)**
<br><br><br><br>

The Association for Computational Linguistics (ACL) is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing.

Transformer Language Models Finally Add Word Frequency Biases

<iframe src="https://app.sli.do/event/uyG7EHDP5trs2LHubv2Gj8/embed/polls/cc529c8d-fcac-4d53-bdef-52e546fd718f" width="300" height="400"></iframe>

Transformer Language Models Handle Word Frequency in Prediction Head

Code search is a task to find programming codes that semantically match the given natural language queries. Even though some of the existing datasets for this task are multilingual on the programming language side, their query data are only in English. In this research, we create a multilingual code search dataset in four natural and four programming languages using a neural machine translation model. Using our dataset, we pre-train and fine-tune the Transformer-based models and then evaluate them on multiple code search test sets. Our results show that the model pre-trained with all natural and programming language data has performed best in most cases. By applying back-translation data filtering to our dataset, we demonstrate that the translation quality affects the model's performance to a certain extent, but the data size matters more.

Constructing Multilingual Code Search Dataset Using Neural Machine Translation

This paper addresses whether masked language models can distinguish (named) entities in internal representations and investigates whether the model's internal representations of the same entity create a sufficiently compact cluster that is distinct from others. The primary contributions of the paper are two-fold. First, we present a novel method for quantitatively assessing the degree to which pre-trained masked language models distinguish between entities in the internal representations. Our approach investigates whether entities belonging to the same object are separated from other embeddings in the embedding space. Second, we conducted experiments, and the findings revealed that internal representations of masked language models were able to distinguish entities from other concepts to a certain degree, even when there was variation in the surrounding context and mentions. (e.g., BERT could distinguish about 70% of the entities from other concepts.) An implication of these results is the possibility that the distinct entities, as reflected in internal representations, are considered evidence of behavior that effectively processes entity strings.

Geometric Locality of Entity Embeddings in Masked Language Models

Pre-trained language models have achieved remarkable results on several NLP tasks. Most of them adopt masked language modeling to learn representations by randomly masking tokens and predicting them based on their context. However, this random selection of tokens to be masked is inefficient to learn some language patterns as it may not consider linguistic information that can be helpful for many NLP tasks, such as multimodal machine translation (MMT). Hence, we propose three novel masking strategies for cross-lingual visual pre-training - more informed visual masking, more informed textual masking, and more informed visual and textual masking - each one focusing on learning different linguistic patterns. We apply them to Vision Translation Language Modelling for video subtitles (Sato et al., 2022) and conduct extensive experiments on the Portuguese-English MMT task. The results show that our masking approaches yield significant improvements over the original random masking strategy for downstream MMT performance. Our models outperform the MMT baseline and we achieve state-of-the-art accuracy (52.70 in terms of BLEU score) on the How2 dataset, indicating that more informed masking helps in acquiring an understanding of specific language structures and has great potential for language understanding.

Choosing What to Mask: More Informed Masking for Multimodal Machine Translation

Chinese modern poetry generation has been a challenging task. One issue is the Chinese word segmentation (CWS) which is critical to comprehend the Chinese language but was not always considered in common tokenization methods. Another is the decoding (sampling) method which may induce repetition and boredom and severely lower the diversity of the generated poetry. To address these issues, we present Lingxi, a diversity-aware Chinese modern poetry generation system. For the CWS issue, we propose a novel framework that incorporates CWS in the tokenization process. The proposed method can achieve a high vocabulary coverage rate with a reasonable vocabulary size. For the decoding method and the diversity issue, we propose a novel sampling algorithm that flattens the high likelihood part of the predicted distribution of the language model to emphasize the comparatively low-likelihood words and increase the diversity of generated poetry. Empirical results show that even when the top 60% of cumulative probability mass of the predicted distribution is flattened, our method achieves comparable or even better performance than baseline sampling methods. Our system is available at http://lingxi.website.

<iframe src="https://app.sli.do/event/qQ87ZDnDYN6cfmCQWHFaok/embed/polls/41e25ef3-986b-4a90-9fce-ae51834bbd9f" width="300" height="400"></iframe>

Lingxi: A Diversity-aware Chinese Modern Poetry Generation System

Spoken Language Understanding (SLU) is one of the core components of a task-oriented dialogue system, which aims to extract the semantic meaning of user queries (e.g., intents and slots). In this work, we introduce OpenSLU, an open-source toolkit to provide a unified, modularized, and extensible toolkit for spoken language understanding. Specifically, OpenSLU unifies 10 SLU models for both single-intent and multi-intent scenarios, which support both non-pretrained and pretrained models simultaneously. Additionally, OpenSLU is highly modularized and extensible by decomposing the model architecture, inference, and learning process into reusable modules, which allows researchers to quickly set up SLU experiments with highly flexible configurations. OpenSLU is implemented based on PyTorch, and released at https://github.com/LightChen233/OpenSLU.

<iframe src="https://app.sli.do/event/qQ87ZDnDYN6cfmCQWHFaok/embed/polls/7ac0a8e3-ea3f-42b9-9c3c-9ed5b493fb1f" width="300" height="400"></iframe>

OpenSLU: A Unified, Modularized, and Extensible Toolkit for Spoken Language Understanding

In this work, we present DIAGRAPH, an open-source graphical dialog flow editor built on the ADVISER toolkit. 
Our goal for this tool is threefold: 
1) To support subject-experts to intuitively create complex and flexible dialog systems,
2) To support rapid prototyping of dialog system behavior, e.g., for research, and 
3) To provide a hands-on test bed for students learning about dialog systems.
To facilitate this, DIAGRAPH aims to provide a clean and intuitive graphical interface for creating dialog systems without requiring any coding knowledge.
Once a dialog graph has been created, it is automatically turned into a dialog system using state of the art language models. This allows for rapid prototyping and testing.
Dialog designers can then distribute a link to their finished dialog system or embed it into a website.
Additionally, to support scientific experiments and data collection, dialog designers can access chat logs. 
Finally, to verify the usability of DIAGRAPH, we performed evaluation with subject-experts who extensively worked with the tool and users testing it for the first time, receiving above average System Usability Scale (SUS) scores from both (82 out 100 and 75 out of 100, respectively).
In this way, we hope DIAGRAPH helps reduce the barrier to entry for creating dialog interactions.

<iframe src="https://app.sli.do/event/qQ87ZDnDYN6cfmCQWHFaok/embed/polls/211b822f-6d56-4e14-8a7f-9422c92eec80" width="300" height="400"></iframe>

DIAGRAPH: An Open-Source Graphic Interface for Dialog Flow Design

In this paper, we present our open-source neural machine translation (NMT) toolkit called "Yet Another Neural Machine Translation Toolkit" abbreviated as YANMTT - https://github.com/prajdabre/yanmtt, which is built on top of the HuggingFace Transformers library. YANMTT focuses on transfer learning and enables easy pre-training and fine-tuning of sequence-to-sequence models at scale. It can be used for training parameter-heavy models with minimal parameter sharing and efficient, lightweight models via heavy parameter sharing. Additionally, it supports parameter-efficient fine-tuning (PEFT) through adapters and prompts. Our toolkit also comes with a user interface that can be used to demonstrate these models and visualize various parts of the model. Apart from these core features, our toolkit also provides other advanced functionalities such as but not limited to document/multi-source NMT, simultaneous NMT, mixtures-of-experts, model compression and continual learning.

<iframe src="https://app.sli.do/event/qQ87ZDnDYN6cfmCQWHFaok/embed/polls/043349e0-2314-4b95-9b5a-dce28e95af83" width="300" height="400"></iframe>

YANMTT: Yet Another Neural Machine Translation Toolkit

Recent years have seen impressive progress in AI-assisted writing, yet the developments in AI-assisted reading are lacking. We propose inline commentary as a natural vehicle for AI-based reading assistance, and present CARE: the first open integrated platform for the study of inline commentary and reading. CARE facilitates data collection for inline commentaries in a commonplace collaborative reading environment, and provides a framework for enhancing reading with NLP-based assistance, such as text classification, generation or question answering. The extensible behavioral logging allows unique insights into the reading and commenting behavior, and flexible configuration makes the platform easy to deploy in new scenarios. To evaluate CARE in action, we apply the platform in a user study dedicated to scholarly peer review. CARE facilitates the data collection and study of inline commentary in NLP, extrinsic evaluation of NLP assistance, and application prototyping. We invite the community to explore and build upon the open source implementation of CARE.

Github Repository: https://github.com/UKPLab/CARE
Public Live Demo: https://care.ukp.informatik.tu-darmstadt.de

<iframe src="https://app.sli.do/event/qQ87ZDnDYN6cfmCQWHFaok/embed/polls/8ffcd0d5-9f71-4cb7-ad41-21590baf58fe" width="300" height="400"></iframe>

CARE: Collaborative AI-Assisted Reading Environment

There are a growing number of table pre-training methods proposed for reasoning over tabular data (e.g., question answering, fact checking, and faithful text generation). However, most existing methods are benchmarked solely on a limited number of datasets, varying in configuration, which leads to a lack of unified, standardized, fair, and comprehensive comparison between methods. This paper presents OpenRT, the first open-source framework for reasoning over tabular data, to reproduce existing table pre-training models for performance comparison and develop new models quickly. We implemented and compared six table pre-training models on four question answering, one fact checking, and one faithful text generation datasets. Moreover, to enable the community to easily construct new table reasoning datasets, we developed TaRAT, an annotation tool which supports multi-person collaborative annotations for various kinds of table reasoning tasks. The researchers are able to deploy the newly-constructed dataset to OpenRT and compare the performances of different baseline systems.

<iframe src="https://app.sli.do/event/qQ87ZDnDYN6cfmCQWHFaok/embed/polls/24e7f2cd-7d8a-4834-9ffd-6beb9f642d59" width="300" height="400"></iframe>

Premium content

Is Anisotropy Inherent to Transformers?

Downloads

Next from ACL 2023

Transformer Language Models Handle Word Frequency in Prediction Head

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES